Inertial navigation gives the position, speed and attitude of an object based on its linear and angular acceleration data. This principle allows a fully autonomous navigation explaining its widespread use within the aero-spatial industry. However, Inertial Measurement Units, systems using this process, are subject to errors growing over the duration of operating time due to numeric integration. This study consisted in modeling such a system and studying its performance using Simulink software, Matlab's graphic programming tool that facilitates the sharing and coding of complex architectures generally intended for embedded equipment. Validating these models through different methods showed good accuracy for low operating times and medium dynamics on inertial measurements.